AI Chat Participant

@weapon Copilot Chat integration with full engagement context awareness.

Prerequisites

Requires GitHub Copilot Chat extension installed and active.

Usage

Open Copilot Chat (Ctrl+Shift+I / Cmd+Shift+I) and type @weapon:

Command	Description	Example
@weapon /analyze	Analyze tool output	@weapon /analyze then paste nmap results
@weapon /suggest	Suggest next pentest actions	@weapon /suggest
@weapon /generate	Generate commands from natural language	@weapon /generate DCSync with impacket
@weapon /report	Summarize findings (direct output, no LLM)	@weapon /report
@weapon /explain	Explain a concept or technique	@weapon /explain Kerberoasting
@weapon <question>	General conversation	@weapon how do I escalate from this user?

How It Works

Context Injection

Every conversation automatically includes:

• System prompt — Pentest-focused instructions and guidelines
• Host context — All discovered hosts with $TARGET = hostname, $RHOST = ip mapping
• User context — All credentials with auth type info and $USER/$LOGIN mapping
• References — Any attached files or selections from the editor

Command Routing

• /analyze, /suggest, /generate, /explain → Full LLM conversation with engagement context
• /report → Direct table generation (no LLM call), outputs hosts + credentials as Markdown tables

Follow-up Suggestions

After each response, contextual follow-up actions are suggested. For example, after /analyze, the participant suggests /suggest and /generate.

Credential Protection

Credential values (passwords, hashes) are never sent to the LLM. The user context builder (buildUserContext() in prompts/userContext.ts) only includes authentication type information (e.g., "password", "hash") and username/domain mappings -- it never includes actual credential values. While AIService.redactCredentials() exists as a utility, the primary protection is that sensitive values are simply never included in the context that is built for the LLM.

Key Files

• src/features/ai/participant.ts — Chat handler, command routing, LLM interaction
• src/features/ai/service.ts — Engagement state access, credential redaction
• src/features/ai/prompts/systemPrompt.ts — System prompt
• src/features/ai/prompts/hostContext.ts — Host context builder
• src/features/ai/prompts/userContext.ts — User context builder